* Switch to the master branch of libp2p
* Fixed having to clear manually
* Fix TTL of bootstrap nodes
* Speed up compilation time
* Update libp2p
* Remove obsolete comment
* Limit number of incoming connections
* Check Endpoint::Listener before checking num_open_connections.incoming
* Maintain at least 1-1/n portion of outgoing connections
* Remove use
* Default incoming_peers_factor to 2
* Use max_incoming_peers and max_outgoing peers to check whether connections should be dropped
* Fix expected_max_peers: reserved peers are not counted in config.max_peers
* typo: fix test
* ok_or → ok_or_else
* Correct the sides of the given and expected args
* Indent } properly
* Convert identation from tabs to spaces.
* Convert identation from tabs to spaces. [2]
* Fix bug and refactoring
- Fix panic in some cases
- Extract the generation as a reusable function
- Improve scoring calculation
- Add tests
- Sort the results
- Fix the case when called with a desired pattern either empty or of 1 char
- Add a 'even more' paranoiac mode
- Add benchmarks
Fix#417
* Fixed compilation for nightly
* Moved chain ops to substrate-service
* moved RPC to susbstrate-service WIP
* Moved RPC to substrate-service
* Version
* Removed redundadnt messages
* substrate CLI
* Removed obsolete files
* Sorted out startup messages
* Pass executable name to CLI
* Simplify cli parsing: remove unnecessary match for Version and help: e.exit() manages both for us already
* Allow passing of IP:PORT as bootnodes, then try to discover its PEER_ID
* Don't connect to ourselves for Kademlia
* Improve log messages
* Manually drop connections
* Kademlia is now 32 seconds
* Fix potential panic
* Fix other potential panic
* No longer pass the endpoint everywhere
* Ping active connections
* Bump rust-libp2p
* Deny connection attempts if we're already connected
* Properly check max_peers against custom connections
* Fix set_until not immediately called
* Clean up unusued connections and stop pinging them
* Disconnect Kad connections after 20 seconds of inactivity
* Make telemetry less susceptible to flakey wifi
* Update readme
* Staging shouldn't autoconnect to telemetry
* Don't try to output more than 1KB of hex to Display
* Better logging of transactions
* Grumbles
* off-by-one
* skeleton of collators object
* awaiting and handling collations. rename `collators` to CollationPool
* add some tests
* add tests
* implement Collators trait for ConsensusNetwork
* plug collators into main polkadot-network
* ignore collator role message
* add a couple more tests
* garbage collection for collations
* extract session-key tracking from consensus
* add local_collations.rs
* finish polish of local_collations
* integrate local_collations into network layer
* introduce API for adding local collations
* mostly finish collator implementation pending service fix
* Specialized network()
* push collations to the network
* grumbles
* substrate-service has custom configuration
* initialize network in collator mode as necessary
* Serve only non-empty Kademlia nodes
* Use the number of custom protos to determine whether to open more
* Add timeout when connecting
* Connect to random peers from the peer store
* Various adjustements
* Typo
* Explicitely connect to bootnodes
* Fix potential overflow
Arkadiy Paronyan